Image Processing Parallel Coding Line Delay
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
As image resolution increases, larger CTU sizes in coding reduce the number of CTU lines in a frame, leading to decreased parallelism in coding and decoding, resulting in increased processing time.
Innovation Solution
An image processing apparatus and method that code and decode coefficient data in parallel across each line of coding tree units, with each line's processing delayed by one coding tree unit relative to the line above it, to maintain parallelism.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If larger CTU size is used in coding, then image resolution and coding efficiency are improved, but the number of CTU lines in a frame decreases, leading to reduced parallelism and increased processing time
Solution Approach 1:
The patent segments the image processing into multiple independent lines of CTUs, where each line can be processed in parallel. By introducing line-by-line processing with controlled delay, the system maintains parallelism even with larger CTU sizes, as each line's processing can overlap with subsequent lines without requiring full frame sequential processing.
Solution Approach 2:
The patent introduces dynamic delay mechanisms where the processing of each CTU line is delayed by a controlled amount (one CTU) relative to the previous line. This dynamic timing adjustment allows the system to maintain optimal parallelism levels adaptively, ensuring that decoding can keep pace with encoding throughout the processing pipeline even when using larger CTU sizes.
2Device complexity
If larger CTU size is used in coding, then fewer CTU lines are present in a frame, but this reduces the number of parallel processing units and increases delay in the parallel pipeline
Solution Approach 1:
The patent applies preliminary action by pre-planning and pre-positioning the delay for each CTU line before processing begins. Each line is scheduled with a predetermined delay (one CTU) relative to the previous line, allowing the parallel processing pipeline to be optimally configured in advance. This prevents timing conflicts and ensures continuous parallel operation without idle waits, reducing overall processing time despite having fewer lines.
3Reliability
If traditional WPP is used with increased CTU size, then coding efficiency is maintained, but parallelism decreases and pipeline delay increases
Solution Approach 1:
The patent changes the timing parameter of CTU line processing by introducing a controlled delay of one CTU between consecutive lines. This parameter modification transforms the traditional WPP approach into a delayed WPP variant that maintains coding efficiency through proper timing synchronization while restoring parallelism. The delay parameter allows each line to be fully processed before the next line begins, preventing data conflicts while enabling overlapping execution of multiple lines.
Data Source
AI summary
Coefficient data related to an image is coded in parallel in each line of coding tree units each including transformation blocks and corresponding to a topmost coding block in a tree structure, and coding of each line is performed so as to be delayed by an amount corresponding to one coding tree unit with respect to coding of a line immediately above the line to be coded. Additionally, coded data obtained by coding coefficient data related to an image is decoded in parallel in each line of coding tree units each including transformation blocks and corresponding to a topmost coding block in a tree structure, and decoding of each line is performed so as to be delayed by an amount corresponding to one coding tree unit with respect to decoding of a line immediately above the line to be decoded.